• 专利附录
  • 专利说明
  • 权利要求
  • 类似技术
  • 同族专利
  • 引用文献
  • 法律状态
  • 优先权
    • 专利摘要:
    Method and device for detecting peak of compressed histogram of pixel values in high-resolution time-of-flight sensors. This document describes a method for detecting the peak of the compressed histogram of pixel values in high-resolution time-of-flight sensors based on the capture of individual photons, as well as a device for detecting the peak of the compressed histogram of pixel values in high-resolution time-of-flight sensors based on the capture of individual photons, which allow precise detection of the histogram peak formed by all the time-of-flight (ToF) measurements obtained for each pixel. (Machine-translation by Google Translate, not legally binding)
    公开号:ES2746248A1
    申请号:ES201830870
    申请日:2018-09-05
    公开日:2020-03-05
    发明作者:Ion Vornicu；Galán Ricardo Carmona；Vázquez Angel Rodríguez
    申请人:Univ De Seviila；Consejo Superior de Investigaciones Cientificas CSIC；
    IPC主号:
    专利说明:

    [0001]
    [0002]
    [0003]
    [0004] OBJECT OF THE INVENTION
    [0005]
    [0006] The object of the invention is framed in the technical field of physics.
    [0007]
    [0008] More specifically, the object of the invention is aimed at time-of-flight sensors (also called ToF for its acronym in English), capable of generating 3D images, by detecting incidence events of a single photon.
    [0009]
    [0010] BACKGROUND OF THE INVENTION
    [0011]
    [0012] The flight time sensors, capable of generating 3D images by detecting incidence events of a single photon, make a direct estimate of the distance at which the objects that make up the scene are located. By means of a pulsed light, the times it takes for the photons to travel through the space separating the light source and the sensor from said objects are measured, while by means of an indirect measurement, different periods of charge integration are established, synchronized with the light pulses. and then the accumulated charges are correlated to obtain an estimate of the flight time. In the direct measure, it is a matter of detecting the moment when the first reflected photons arrive. Given that the arrival of the photons to the sensor is subject to the statistics of the transmission and reflection of light in the optics of the system, the efficiency in the detection of photons from the avalanche diode and the probability that such an avalanche will start, is You can establish that it is a statistical phenomenon that is also affected by spurious avalanches generated by other mechanisms and by the incidence of photons from background lighting . In order to try to separate the relevant information from noise, it is necessary to take an important number of measurements and carry out an appropriate filtering of them to improve the signal-to-noise ratio.
    [0013]
    [0014] Time-of-flight CMOS image sensors based on individual photon detection using, for example, avalanche diodes (SPAD), requires filtering of the readings that are obtained from the pixels. To obtain an accurate estimate of depth, taking into account that photon detection is a statistical phenomenon subject to quantum efficiency in photon detection and the probability of avalanche generation, it is necessary to carry out a certain number of measurements. Furthermore, it becomes necessary to effectively separate the data that actually gives a ToF measure of the uncorrelated noise corresponding to the dark counts and the backlighting, which appear superimposed on the pixel value.
    [0015]
    [0016] A very useful tool for this separation is the construction of a histogram from a number of readings from each of the pixels. Future generations of time-of-flight sensors will have higher spatial and temporal resolution, along with a greater dynamic range and higher refresh rate. Under these circumstances, storing the entire histogram for each pixel becomes virtually impossible.
    [0017]
    [0018] The techniques known in the state of the art are based on algorithms such as the STG technique (initials of its Anglo-Saxon denomination Scanning Time Gated) which consists of sweeping a small time gate throughout the dynamic range and searching for ToF information at each step by constructing partial histograms. The number of steps is proportional to the dynamic range (DR for its Anglo-Saxon name dynamic range) divided by the time window (TG for its Anglo-Saxon name time gate) and inversely proportional to the frame rate.
    [0019]
    [0020] The technique for displaced optical light radiation consists of scanning the entire dynamic range where scanning is performed by shifting the light pulse instead of the acquisition TG. A histogram is produced at each step of the scan. All the histograms are combined in a processor to obtain the final histogram.
    [0021]
    [0022] Another known technique is complete histogram construction which involves direct histogram construction based on ToF samples.
    [0023]
    [0024] Likewise, there is evidence of document US20170052065A1 in which the STG ( Scanning Time Gated ) technique is described, in which during a time window (TG) the entire dynamic range (DR) is scanned, constructing a histogram thickness (CH) for each position of the TG, that is, there are a number of histograms NCH = DR / TG; which is equivalent to practically dividing the entire histogram corresponding to the entire dynamic range into NCH partitions, in fact a larger NCH is required due to overlapping consecutive TGs to avoid edge effects. The longer the dynamic range, the higher the NCH, which means a lower frame rate. This happens because all partial histograms must be evaluated to decide where the true ToF data is located. However, this STG methodology is susceptible to false positive detections, especially for low ToF Signal Noise Ratio . Also, in STG, the TG is placed at the true ToF location and a final histogram is constructed from more measurements. Without the time window overlap, STG-based techniques are affected by edge effects.
    [0025]
    [0026] In US20170003382A1 a method of preparing histograms of a sensor signal from a sensor array, in particular proximity sensors, and the corresponding device are detailed. More specifically, a scan like the one cited in US20170052065A1 is detailed, but instead of scanning the time window over the entire dynamic range, scanning is carried out by displacing the optical pulse; This methodology described in US20170003382A1 has an additional inherent limitation, and is that related to the frame rate.
    [0027]
    [0028] EP2469301A1 describes how to generate a 3D representation of a scene at very high speed using a progressive scan technique. In addition to comprising two phases, i.e. coarse and fine, the ToF estimation based on burst mode acquisition involving parallel detectors coupled to light diffusers, 10 GHz clock references, ultrashort laser pulses and averaging of weather. The approach is based on progressive exploration; With the approximate distance already known, data processing circuits can determine which pulse a given photon count results from, although this configuration is normally affected by ambiguity errors. Multiple SPADs are used in parallel with the diffuser to avoid detector dead time.
    [0029]
    [0030] In US20150041625A1 a histogram construction method is detailed using a direct approach (calling the histogram direct), without involving any processing of the histogram; as the direct implementation of the Full histogram is not suitable for large arrays, high resolution ToF and wide dynamic range (~ 300ns) due to high memory requirements. Although a data compression is detailed in this document, it always refers to the complete construction of complete histograms, which is highly inefficient increasing the limitations in the dynamic range of ToF, that is, the area of the histogram scales by 2AN, where N is the number of bits ToF. Furthermore, each channel of the histogram has to be implemented by a counter for detection of multiple events, which is even more inefficient. An alternative implementation referred to in US20150041625A1 is performed with a parallel thermometric code to binary converter and adder cells. In both cases, this approach cannot be applied for a high dynamic range due to memory size limitations.
    [0031]
    [0032] US006504954B1 details both a method and an apparatus of specification of linear histogram at closed intervals by parts; that is, it describes an invention related to systems and methods that implement piecewise linear histogram specification processing and is employed in infrared video processing electronics (FLIR); more specifically the object of the described invention is specifically designed for 2D video and image processing. US006504954B1 works between an internal high dynamic range representation of the video signal and a lower dynamic range representation suitable for displaying on a monochrome monitor, having as main application contrast enhancement and dynamic range conversion in the processors of video. The system and methods described in US006504954B1 use a span line transfer function coupled with a collection of thick histograms to reduce bandwidth compared to the conventional histogram method; which implies that a closed loop algorithm is also required to achieve a specified thick histogram, having a number of trays in the thick histogram equal to the number of segment linearized segments. The stretch linearization described in US006504954B1 is used to reduce the digital input data so that a thick histogram can be constructed, which has the same number of intervals as the number of segment linearized segments, that is, fewer intervals than the conventional histogram method; however, the construction of the histogram is conventional. Compression of the input data is based on the span linearization method.
    [0033] DESCRIPTION OF THE INVENTION
    [0034]
    [0035] The object of the invention, both in its aspect related to the method and the aspect related to the system, is conceived for the practical realization of this filtering, specifically the detection of the histogram peak formed by all the flight time measurements obtained each pixel. It is a compressed histogram constructor that operates with a reduced amount of memory using a coarse-fine approximation that allows to discard the memorization of those areas of the histogram that do not contain relevant information. As a result of this simplification, it is possible to implement a processing channel of the readings generated in each pixel in real time, which would allow obtaining the filtered 3D image in real time. Therefore, the object of the proposed invention is based on a totally different approach of constructing a compressed histogram like those known in the prior art; More specifically, the proposed invention is based on the change of histograms that zooms directly into the complete histogram and implies that it is not necessary to use the scan, which is a great advantage from the point of view of speed; for example, instead of acquiring N coarse histograms CH, p. ex. 81, only one histogram is required. Furthermore, the object of the invention is more precise and much faster than, for example, those known techniques based on photon counting in scanned TGs; by way of example, the object of the invention requires only one CH, regardless of the dynamic range.
    [0036]
    [0037] The object of the invention contemplates the possibility of working with flight time readings of up to 15 bits, which allows it to be applied to flight time image sensors of future generations in which an increase in temporal resolution can be anticipated and spatial, and an increase in the dynamic range and refresh rate of the image.
    [0038]
    [0039] The object of the invention allows, taking into account that most of the information contained in the histogram corresponds to the noise level, to efficiently store only those data, considered relevant, necessary for the calculation of flight time (ToF by its acronym in English).
    [0040]
    [0041] The subject of the invention employs fine conversion by simply zooming in the entire histogram, centered on the ToF data. This technique allows the construction of, for example, only 2 histograms instead of at least 82 for the sweep focus (81 CH 1 final histogram). Consequently, an increase in the dynamic range does not imply a decrease in the frame rate as it occurs in all known scanning methods, allowing the variable number of ToF bits using the same hardware: which is a great advantage for image readers. adaptive frame rate. The object of the invention has special application in implementations with low signal-to-noise ratio, which can occur in LiDAR applications with high backlight.
    [0042]
    [0043] The method object of a first aspect of the invention allows the extraction of the most repeated value in a histogram (formed by the estimates of the time of flight of light measured by a pixel based on SPADs. This object of the invention It exploits the fact that the unrelated noise is evenly distributed throughout the histogram and that the relevant information is concentrated around the peak value, so there is no need to store the entire histogram.
    [0044]
    [0045] To do this, we proceed to move the values to the most interesting area of the histogram, to which we applied the analysis (detection of the most frequent value) that was previously performed on the entire histogram. In other words, a histogram of the data of each pixel in the intermediate frames can be used, moved to the area with the highest probability of photon detection, which requires a memory that is 128 times smaller than what would be necessary. to store the entire histogram. In fact, a 2E + 08 bit memory would be enough to process a histogram that originally contained 2E + 15 segments while the overall refresh rate would drop by only half.
    [0046]
    [0047] In a second aspect of the object of the invention there is a device referring to a possible hardware implementation of the method of the first aspect of the invention, such as an implementation on a chip such as an FPGA or a specific circuit. The proposed implementation as this second aspect of the invention can be given by the incorporation of a circuit to extract the ToF value from the histogram in real time, just as at the end of the acquisition phase, the ToF is available immediately, without require additional processing.
    [0048]
    [0049] DESCRIPTION OF THE DRAWINGS
    [0050] To complement the description that is being made and in order to help a better understanding of the characteristics of the invention, according to a preferred example of practical embodiment thereof, a set of drawings is included as an integral part of said description. where, by way of illustration and not limitation, the following has been represented:
    [0051]
    [0052] Figure 1.- Shows the three types of histograms that are discussed in the text: the complete histogram (6), the thick histogram (7) and the fine histogram (10) as well as a zoom of the complete histogram (6). These histograms contain measurements obtained by a single pixel.
    [0053]
    [0054] Figure 2.- Shows a flow chart of the algorithm with a loop for detecting the position of the peak of the thick histogram (7) on the left and a loop for detecting the position of the peak of the fine histogram (10) at the left. right.
    [0055]
    [0056] Figure 3.- Shows a diagram showing the different blocks of the physical implementation of the object of the invention defined as a second aspect thereof. The diagram shown corresponds to Nv = 15, Nh = 10 and Nb = 8.
    [0057]
    [0058] Figure 4.- Shows a schematic diagram of the internal structure of the peak position detector.
    [0059]
    [0060] Figure 5.- Shows a diagram representing the internal structure of the digital filter.
    [0061]
    [0062] PREFERRED EMBODIMENT OF THE INVENTION
    [0063]
    [0064] In a possible preferred embodiment of the object of the invention, it is assumed that a complete histogram (6) (such as the one shown in Figure 1) formed by a certain number of estimates of the ToF flight time in a pixel with A SPAD comprises two fundamental elements:
    [0065]
    [0066] • A Gauss bell (4) that encodes the ToF value at the position of the maximum value of said Gauss bell (4), and the jitter of the combined SPAD and TDC, in their standard deviation.
    [0067] • A noise threshold (5) that contains all the measurements that are not correlated with the arrival of the photons from the pulsed source reflected by the objects in the scene.
    [0068]
    [0069] It can be understood that it is not necessary to store all the samples that define this noise threshold (5), since they contain redundant information about the number of occasions when a spurious avalanche is triggered. In fact, only a small number of bars in the histogram contain relevant information.
    [0070]
    [0071] Let us consider that 2Nb bars are sufficient for this task, regardless of the number of bits used to encode the pixel value ( Np). The key is to find a way to zoom the entire histogram so that the aforementioned Gauss bell (4) is centered on a histogram of only 2Nb bars.
    [0072]
    [0073] The proposed method consists of the generation of a thick histogram (7) and a fine histogram (10), each preferably of 2Nb bars, for which we will need a memory of 2Nb ■ Nh bits. In practice, if we take Nb = 8, we will be able to calculate histograms of ToF on-the-fly of sensors that generate pixel values with an Np of up to 15 bits.
    [0074]
    [0075] The procedure we are going to follow is described by the block diagram in Figure 2. First, we generate a rough (coarse) approximation of the peak position by constructing a coarse histogram (7) with the most significant Nb bits of the value of the pixel, as can be seen in figure 1. This thick histogram (7) is obtained by filtering the ToF measurements, eliminating the least significant ( Np - Nb) bits and keeping only the most significant Nb bits. From this thick histogram (7) the peak position, b Thick, is obtained by detecting the ToF measurement that is most repeated. This procedure is described by the loop on the left of the diagram in Figure 2.
    [0076]
    [0077] Next, the fine histogram (10) is obtained, which is obtained by making a second capture with the filtered pixel values with TH + and TH- thresholds that refer to the entire histogram (6), with values encoded in Nb bits. This fine histogram (10) is obtained in a second capture. From a thick bM value, calculates an upper threshold (8) and a lower threshold (9) that can be identified in figure 1:
    [0078] TH + = (2 Np ~ Nb) bM, coarse 2Nb ~ 1 - offset
    [0079] TH_ = ( 2Np ~ Nb) bM, g coarse - 2 N b ~ 1 - offset
    [0080] The thresholds (8.9), are referred to the range of the complete histogram (6). The thresholds (8.9) allow determining if the pixel values that are being captured for the generation of the fine histogram (10) are within a 2Nb level window centered around the maximum of the complete histogram (6). If the pixel value is not between the thresholds (8.9), it is discarded, if it is between the thresholds (8.9), a pixel value shift is performed in order to accommodate these values in a fine histogram (10) with 2Nb levels, that is, whose bars are in positions encoded by only Nb bits. This displacement is given by:
    [0081]
    [0082] A = 2Nb floor
    [0083]
    [0084] This fine histogram (10) will also have a maximum value, the position of which is min. In order to correctly determine the position of said maximum, this value of bMñno, which is represented in a range of Nb bits, must be shifted to the range described by the original Np bits of the complete histogram (6), so that the position of the value of the peak corresponding to the precise measurement of the ToF is finally:
    [0085]
    [0086] Bm = ^ M, fine A
    [0087]
    [0088] In this way we will have obtained the most repeated value of a histogram of values encoded by Np bits, with the memory resources necessary to obtain a histogram of values encoded by only Nb bits. We will use this memory to obtain the thick histogram (7) and then we proceed to overwrite it to obtain the fine histogram (10). In this way we will use a memory 2Np ~ Nb times smaller.
    [0089] The physical implementation of the object of the invention consists of the blocks exposed in figure 3. In this diagram, Np = 15, Nh = 10 and Nb = 8. The main blocks of the diagram are: a deserializer (11) preferably type SIPO of Np bits of serial input and parallel output, a register (12) preferably of the PIPO type of parallel input and output of Np bits, a digital filter (13), a multiplexer (14) with preferably two Nb bit inputs, a memory ( 15) preferably static RAM and Nh x 2Nb bits for storing the thick histogram (7) and the fine histogram (10), a register (16) of Nh bits with automatic increase of 1 unit, a peak detection circuit (17), an algorithmic block destined to perform addition, subtraction, multiplication and division, grouped into blocks Alg 1 (18), Alg2 (19) and Alg3 (20), and two memories (21,22) of Np bits respectively, to store the value of offset A and the final value obtained for the ToF of the pixel.
    [0090]
    [0091] Before processing a new trame the RST_FR signal, with this memory (15) and the detection circuit the peak (17) is reset is activated. After this, the value of the first pixel is sent serially bit by bit to the deserializer (11). Every Np CLK clock periods, the content of the deserializer (11) is transmitted to the register (12) and therefore that value has Np CLK periods to be incorporated into the thick histogram (7) or fine histogram (10) according to the phase of the algorithm, in the bar where it corresponds.
    [0092]
    [0093] The first phase in calculating ToF is the construction of the thick histogram. To achieve this, an AC input is first selected from the inputs in the multiplexer (14), and then read the contents of the memory address (15) indicated by the ADDR value, copy it to register (16) and increase its value by 1 unit. If the resulting value is greater than that currently stored in the peak detection circuit (17), this resulting value is saved as the largest maximum value present in the thick histogram (7). Once the last pixel value is resolved, the thick histogram (7) is complete. From the position of the peak value found, the TH + and TH_ values that are used within the digital filter (13) are generated using the Alg 1 arithmetic block (18). We also proceed to obtain the value of A that is stored in the memory (21) shown in Figure 3. It should be mentioned that the division performed in the algorithmic block (19) has been implemented sequentially in order to reduce the module area.
    [0094] With the calculation of the TH +, TH_ and A values, which is performed in Np x Tclk clock cycles, the memory (15) is reset, and the peak detector circuit (17) that has an additional first register (24) that stores the current peak of the thick histogram (7); a second additional register (25) containing the code in which the current peak of the thick histogram (7) is located; and a third additional register (26) destined to retain the code in which the final peak of the thick histogram (7) is located during the construction of the fine histogram (10).
    [0095]
    [0096] The second phase in calculating ToF is the construction of the fine histogram. When the construction of the thick histogram (7) is finished, the thick bM value is saved in a third additional register (26) with PNoCc output. This third additional record (26) is overwritten at the end of the construction of the thick histogram (7). Following the reset of the memory (15) and the peak detection circuit (17), the operation of the digital filter (13) is enabled, and the SA input on the multiplexer (14) and a second acquisition is started . The digital filter (13) allows you to discard those pixel values that are below TH_ and above TH +. Furthermore, since the pixel values being read are described by Np bits, it is necessary to subtract the value of A, which is done in a subtraction circuit (27) that can be seen in figure 5 that represents the internal structure of the digital filter (13), and keep only Nb bits, which after all will be the only ones that can be different from noise. This procedure can be seen in the loop to the right of the flow chart in Figure 2.
    [0097]
    [0098] Finally, the value of the flight time, which corresponds to the position of the peak of the fine histogram (10), is obtained by adding the value of A to the value found in the second additional register (25), which is a register with PNoCf output, which which takes place in an adder circuit (23) like the one shown in figure 3.
    权利要求:
    Claims (5)
    [1]
    1. Peak detection method of the compressed histogram of pixel values in high-resolution time-of-flight sensors, the method being characterized by comprising:
    to. delimit an area of a complete histogram (6),
    b. generate a thick histogram (7), removing the least significant ( Np - N b) bits and keeping only the most significant Nb bits of the pixel value,
    c. obtain from the thick histogram (7) a position corresponding to a pixel value that has been repeated the most times, bThick. d. calculate an offset value A that is given by:
    TO
    (2 wp Nb) V coarse + 2Nb 1 - offset
    = 2Nb jfloor
    2Nb (2 Np ~ Nb) b ~ 1 - offset
    mod M, coarse + 2 Nb
    2Nb
    and. calculate an upper threshold value (8) TH + and a lower threshold value (9) TH _ according to:
    TH + = ( 2Nv ~ Nb) bM> coarse + 2 "* -1 - offset
    TH _ = ( 2Np ~ Nb) bMy thickness - 2Nb ~ 1 - phase shift
    F. obtain a fine histogram (10) with values encoded in Nb bits by shifting the acquired values that are encoded in Np bits, selecting by means of a digital filter (13) only those pixel values that are within a window of 2Nb levels around the maximum of the Complete histogram (6) delimited by the upper (8) TH + and lower (9) TH_ thresholds.
    g. calculate the ToF value, B M, from the fine histogram (10) by moving the value of bMñno to the range described by the original Np bits, using:
    Bm = ^ M, fine A
    [2]
    2. Device for detecting the peak of the compressed histogram of the pixel values in high-resolution time-of-flight sensors, characterized in that it is configured to carry out the method of claim 1 and that it comprises:
    to. a deserializer (11) preferably type SIPO of Np bits of serial input and parallel output,
    b. a register (12) preferably type PIPO of parallel input and output of Np bits,
    c. a digital filter (13), intended to select and discard pixel values based on threshold values in such a way that only those comprised between said threshold values are selected, d. a multiplexer (14) with preferably two Nb- bit inputs, e. a memory (15) for storing the fine histogram (10) and the thick histogram (7),
    F. a register (16) of Nh bits with automatic increase of 1 unit, g. a peak detection circuit (17),
    h. an algorithmic block intended for addition, subtraction, multiplication and division, and comprising a series of blocks Alg1 (18), Alg2 (19) and Alg3 (20), and
    i. two memories (21, 22) of Np bits respectively to store the values of A and the final value obtained for the ToF of the pixel.
    [3]
    3. Device according to claim 2, characterized in that the digital filter (13) comprises a subtraction circuit (27) intended to subtract the value of A, from the pixel values described by Np bits.
    [4]
    4. Device according to claim 2, characterized in that the peak detection circuit (17) comprises three internal registers:
    to. a first additional register (24) to save a resulting value as the largest value present in the thick histogram (7) and the fine histogram (10) according to the phase of the algorithm;
    b. a second additional register (25) to store the position of the largest corresponding value; Y
    c. a third additional register (26) to store the position of the final maximum value in the thick histogram (7);
    where the position of the final maximum value in the fine histogram (10) is stored in the second additional register (25).
    [5]
    5. Device according to claim 2, characterized in that the memory (15) is a static RAM type with Nh x 2Nb bits.
    类似技术:
    公开号 | 公开日 | 专利标题
    CA2950907C|2020-01-07|Linear mode computational sensing ladar
    FR3043797A1|2017-05-19|
    Van Dam et al.2011|A practical method for depth of interaction determination in monolithic scintillator PET detectors
    US20160306045A1|2016-10-20|A method for driving a time-of-flight system
    CN108291961A|2018-07-17|Solid-state imaging apparatus, Distnace determination device and method for measuring distance
    US20140347553A1|2014-11-27|Imaging devices with light sources for reduced shadow, controllers and methods
    US8908063B2|2014-12-09|Method and apparatus for a time-of-flight sensor with charge storage
    ES2746248B2|2021-08-05|METHOD AND DEVICE FOR DETECTING THE PEAK OF THE COMPRESSED HISTOGRAM OF PIXEL VALUES IN HIGH RESOLUTION TIME-OF-FLIGHT SENSORS
    WO2011051529A2|2011-05-05|Method and electronic circuit for reading the signals generated by one or more pixelated sensors
    Vornicu et al.2018|Compact real-time inter-frame histogram builder for 15-bits high-speed ToF-imagers based on single-photon detection
    US20200233068A1|2020-07-23|Digital pixels and operating methods thereof
    Peng et al.2020|Photon-efficient 3d imaging with a non-local neural network
    Martel et al.2017|High-speed depth from focus on a programmable vision chip using a focus tunable lens
    WO2018180660A1|2018-10-04|Solid-state imaging device
    US20200092540A1|2020-03-19|Detection device and detection method using avalanche diode array and calibration matrix generating method thereof
    Boksenberg et al.1986|Interpolative centroiding in CCD-based image photon counting systems
    JP2020524258A|2020-08-13|Systolic processor system for optical ranging system
    US10353087B1|2019-07-16|Coincidence circuit with splitter
    CN107184223B|2020-06-23|Radiation detector and TOF-PET device provided with same
    WO2020253968A1|2020-12-24|Direct time-of-flight depth sensor architecture and method for operating of such a sensor
    US20210072360A1|2021-03-11|Photon detecting 3d imaging sensor device
    Li et al.2018|Fast Monolithic Detector Calibration Method-a simulation study
    Valastyán et al.2012|Novel time over threshold based readout method for MRI compatible small animal PET detector
    WO2020261278A1|2020-12-30|Digital readout enabling 2d and 3d analysis for silicon photo multiplier
    Colmeiro et al.2021|Event Localization in Continuous Crystal Scintillation Cameras using Distribution Matching Neural Networks
    同族专利:
    公开号 | 公开日
    ES2746248B2|2021-08-05|
    WO2020049203A1|2020-03-12|
    引用文献:
    公开号 | 申请日 | 公开日 | 申请人 | 专利标题
    US20030218129A1|2002-02-14|2003-11-27|Bruker Daltonik Gmbh|High resolution detection for time-of-flight mass|
    US20070278409A1|2006-06-06|2007-12-06|General Electric Company|Methods and apparatus for pet time of flight|
    JP2016161438A|2015-03-03|2016-09-05|株式会社デンソー|Arithmetic unit|
    EP3367131A1|2017-02-27|2018-08-29|STMicroelectronics Limited|A higher pixel density histogram time of flight sensor|
    US6504954B1|1999-02-05|2003-01-07|Raytheon Company|Closed loop piecewise-linear histogram specification method and apparatus|
    EP2469301A1|2010-12-23|2012-06-27|André Borowski|Methods and devices for generating a representation of a 3D scene at very high speed|
    GB2520232A|2013-08-06|2015-05-20|Univ Edinburgh|Multiple Event Time to Digital Converter|
    FR3038433A1|2015-07-02|2017-01-06|Stmicroelectronics Sas|METHOD FOR PRODUCING HISTOGRAMS OF A SIGNAL OF SENSORS FROM A SENSOR MATRIX, PARTICULARLY OF PROXIMITY, AND CORRESPONDING DEVICE|
    US10620300B2|2015-08-20|2020-04-14|Apple Inc.|SPAD array with gated histogram construction|
    GB201622429D0|2016-12-30|2017-02-15|Univ Court Of The Univ Of Edinburgh The|Photon sensor apparatus|US20210302550A1|2020-03-30|2021-09-30|StmicroelectronicsLimited|Time to digital converter|
    法律状态:
    2020-03-05| BA2A| Patent application published|Ref document number: 2746248 Country of ref document: ES Kind code of ref document: A1 Effective date: 20200305 |
    2021-08-05| FG2A| Definitive protection|Ref document number: 2746248 Country of ref document: ES Kind code of ref document: B2 Effective date: 20210805 |
    优先权:
    申请号 | 申请日 | 专利标题
    ES201830870A|ES2746248B2|2018-09-05|2018-09-05|METHOD AND DEVICE FOR DETECTING THE PEAK OF THE COMPRESSED HISTOGRAM OF PIXEL VALUES IN HIGH RESOLUTION TIME-OF-FLIGHT SENSORS|ES201830870A| ES2746248B2|2018-09-05|2018-09-05|METHOD AND DEVICE FOR DETECTING THE PEAK OF THE COMPRESSED HISTOGRAM OF PIXEL VALUES IN HIGH RESOLUTION TIME-OF-FLIGHT SENSORS|
    PCT/ES2019/070586| WO2020049203A1|2018-09-05|2019-09-04|Method and device for detecting the peak of the compressed histogram of pixel values in high-resolution time-of-flight sensors|
    [返回顶部]